An experiment based on a two-by-two factorial design was conducted to test hypotheses involving the relationship of four supervisory styles to aggressive feelings and actions of subordinates. The supervisory styles were arrayed on two continuums: the close versus general and the punitive versus non-punitive. Close as compared with general supervision produced significant increases in aggressive feelings toward the supervisor and in indirect aggression toward the supervisor through lowered productivity, an insignificant increase in verbal aggression toward the supervisor, and an increase of borderline significance in aggressive feelings toward co-workers. Punitive as compared with non-punitive supervision produced significant increases in indirect aggression through lowered productivity and in verbal aggression, but no significant increases in aggression toward co-workers. The relationship between close supervision and aggresive feelings appears to be mediated by the self-esteem of the subordinate; an increase in aggresive feelings occurred only in subjects having low self-esteem. Finally, the combined effect of the close and punitive supervision dimensions, for both aggressive feelings and indirect aggression, was not a simple function, but was less than would be predicted on the basis of additive assumptions.